This document discusses using Watson Natural Language Classifier to tag ideas from an M-CAFE dataset with topics. 106 ideas were split randomly into a training set of 86 tagged ideas and a test set of 20 untagged ideas. Watson was trained on the training set and tested on the test set, achieving an accuracy of 80%. Examples of correctly and incorrectly classified ideas are provided. Questions are also included about how the classifier is trained and whether unspervised classification is available.